OSMlanduse

OSMlanduse provides European Union land use information at 10 m resolution by fusing OpenStreetMap labels and Sentinel-2 remote sensing data for continental-scale land use mapping.

Key Features:

• Open Data Integration: The method relies solely on openly available data streams, eliminating training data acquisition or the need for additional expert knowledge.
• Large-Area Fusion: The map represents the first successful large-area fusion of OpenStreetMap and Copernicus data at a spatial resolution of 10 m or finer, applicable globally.
• High Overall Accuracy: The overall map accuracy is 89%, with class-specific accuracies ranging from 77% to 99%.
• Direct Label Mapping: OpenStreetMap tags were translated into land use labels, directly mapping 61.8% of the Union's area.
• Web Map Service: The dataset enables exploration and analysis of land use patterns across the EU through an associated web map service.

Scientific Applications:

• Land Use Mapping: The dataset supports creation of contiguous land use maps for large geographic areas using volunteered geographic information and satellite imagery.
• Remote Sensing Classification: The approach addresses varying label noise and feature space quality in continental-scale land use classification from multispectral data.

Methodology:

OpenStreetMap tags were translated into land use labels to directly map 61.8% of the EU area; these labels served as training data for country-specific deep learning convolutional neural networks applied to Sentinel-2 feature space composites of 2020 at 10 m resolution to predict land use in remaining areas.